Defect Detection in Pharma Pills Using Image Processing

Abstract

In this work, methods for finding defects in pharma pills are proposed. Here pills are classified as tablets and capsules, further tablets are classified based on their shapes as oval shaped and round shaped tablets. Capsules are classified based on their colors in this three colors are considered that is red, green and blue double colored capsules like white-blue and white-brown are considered. While packing there may be some visible defects in the pills. This will vary the dosage of pills, manual inspection would be too tedious and less accurate so here some methods to identify these defects are proposed. Defects such as variation in count of tablets, cracks, breaks or variations in the size and shapes of tablets are considered. In capsules absence of capsule, variation in the size and shape of the capsules or presence of any other colored capsules are considered. These methods successfully detect number of non-defected tablets and number of defected tablets, and hence the non-defected tablets can be reused and defected tablets can be discarded.